Using cognitive psychology to understand GPT-like models needs to extend beyond human biases

This letter argues that using cognitive psychology to understand GPT-like models needs to extend beyond human biases. While Binz and Schulz used cognitive psychology to understand GPT-3, the approach should be expanded to consider a wider range of cognitive phenomena. The authors highlight how cognitive network science reveals bias in GPT models mirroring math anxiety in high-school students, and call for lessons from the study of natural cognition to be applied more broadly to artificial intelligence systems.